You don't need to parametrize Mg2+. Every force field already has parameters for it, and the parameters for ATP should not depend on Mg2+ at all, anyway, as doing so would create non-transferable parameters.

Dear Justin Lemkul ,

I didn't get what you said, Mg2+ is coordinated to beta and gamma oxygens of ATP, it's a part of the system, how can I generate topology without considering Mg2+? There will be bond angles and lengths associated with Mg2+ & beta, gamma oxygens of ATP.

Regards,

Raghav

Your proposed parametrization strategy lacks transferability and generality. You would end up with a set of charges, bonded parameters, etc. that are specific to ATP and Mg2+, but only in an isolated complex. If they're bound to a protein, are they still valid? Moreover, no additive force field parametrizes species in this way. Parameters are obtained for small molecules and ions based on known, experimental data. If parametrized in consistent ways, using consistent levels of theory, etc. then you can safely combine them. ATP and Mg2+ will be held together by a very strong nonbonded electrostatic attraction. You don't need to try to parametrize bonds between them (again, you would get parameters for the specific geometry you supply, which may not be correct when in an actual biological system).

Or, you could save yourself a lot of trouble and use a force field that already covers these species natively (AMBER and CHARMM already have everything you need, for instance).

Dear Justin Lemkul ,

Thanks a lot, your proposed parametrization strategy worked!

I have another question now which is bugging me, I have been using Acpype for generating topologies for ligands and I used lig_GMX_OPLS.top file generated by acpype but when I went through the lig_GMX.itp file I didn't see any opls parameters in it but the top line of the topology says #include "ffoplsaa.itp".

Does this mean that I have mixed amber and opls force fields? because acpype use antechamber which is based on GAFF but the topology file says #include "ffoplsaa.itp". I have attached the topology and itp file.

Regards,

Raghav

Apparently the topology claims compatibility with OPLS-AA, but uses totally custom atom types. I have no experience with these tools so I cannot comment on the suitability of this approach.